Ontology Matching Using Weighted Graphs

In this paper, we present a new way towards ontology matching. Using the graph representation for ontologies and schemas we proceed to calculate the weights for each node of the graph using the lexical similarity of the ancestors. With the guiding intuition that, if the parent nodes match then their children are likely to match as well. This simple observation helps on to build a fast and efficient algorithm for matching different graphs (which represent ontology or schema). Since the algorithm is very fast it can be used as a quickly and dirty method to do initial matching of a large dataset and then drill down to the exact match with other algorithms. The algorithm is not dependent on the method used for calculating the lexical similarity so the best lexical analysis can be used to derive the weights. Once the weights are in place we can calculate the matching in just a single traversal of the graphs. No other algorithm that we know of can give such fast response time.